The Labour Share in G20 Economies

International Labour Organization

Organisation for Economic Co-operation and Development
with contributions from
International Monetary Fund and
World Bank Group

Report prepared for the G20 Employment Working Group

Antalya, Turkey, 26-27 February 2015

1
Introduction

National income is the sum of all income available to the residents of a given country in
a given year. The division of national income between labour and capital is called the
functional distribution of income. The labour income share (or labour share) is the part
of national income allocated to labour compensation, while the capital share is the part
of national income going to capital. A falling labour share often reflects more rapid
growth in labour productivity than in average labour compensation, and an increase in
returns to capital relative to labour. A detailed description of the labour share and how it
is measured is presented in Annex A.

Labour shares have long been considered stable and therefore attracted little attention
from research and policy discussions. Yet, in recent years, a growing body of evidence
suggests that labour shares have seen a secular downward trend with important negative
consequences. For instance, with declining labour shares, improvements in
macroeconomic performance may not translate into commensurate improvements in
personal incomes of households (Atkinson 2009). And data shows that over time and
across many countries, a higher capital share is associated with higher inequality in the
personal distribution of income (Piketty 2013). A declining labour share can also have
political consequences if it erodes support for market-oriented economic policies or for
globalization more broadly. Importantly, trends in labour shares negatively affect the
main macroeconomic aggregates, namely household consumption, private sector
investment, net exports and government consumption (ILO 2012; Wolf 2014).

This paper reviews recent trends in the labour share in G20 countries (and over a long
period of time in a few) and discusses possible causes of the observed trends. It then
explores linkages between the labour income share and the main components of
aggregate demand. Other critical issues such as the growth and employment impacts of
the labour share and policy implications are raised in the paper Strengthening the link
between employment and growth submitted to the EWG meeting of 26-28 February
2015 and will be discussed in more detail in a follow-up paper for the third EWG
meeting (23-25 July 2015).

1. Recent trends in G20 countries

(a) Labour shares

At least until the 1980s, a stable labour income share was accepted as a stylized fact of
economic growth.1 Over the past decades, however, this conventional wisdom has been

1 These empirical findings date back to the early twentieth century, when Arthur Bowley first observed such
regularity using British data from the nineteenth and twentieth centuries and formulated Bowleys Law.
Paul Douglas made a similar finding regarding the labour share in the United States, and developed,
together with the mathematician Charles Cobb, the famous CobbDouglas production function, which
simplifies economic modelling by assuming that the functional income distribution between labour and
capital always remains constant (see Mankiw, 2003). Keynes described this empirical constancy as a bit of

2
challenged by the empirical evidence, which indicates a downward trend for the labour
share in many of the countries for which data are available. The OECD (2012) has
observed, for example, that over the period from 1990 to 2009 the share of labour
compensation in national income declined in 26 out of 30 advanced countries for which
data were available, and calculated that the median (adjusted) labour share of national
income across these countries fell from 66.1 per cent to 61.7 per cent. A more recent
OECD calculation finds that the average adjusted labour share in G20 countries went
down by about 0.3 percentage points per year between 1980 and the late 2000s.2 Similar
downward trends have been observed by other international institutions (IMF, 2007;
European Commission, 2007; BIS, 2006; ILO, 2012).

In the case of emerging and developing economies, the evidence appears to be more
mixed and somewhat ambiguous. Nonetheless, the ILO finds that in many emerging
and developing countries the decline in the labour income share is even more
pronounced than in advanced economies, with considerable declines in Asia and North
Africa and more stable but still declining wage shares in Latin America (ILO-IILS, 2011).
Two recent global reviews of labour income shares also documented significant declines
in some of the large labour-abundant emerging countries (Karabarbounis, 2013), while
also pointing out the greater fluctuations and oscillations in developing countries
(Guerriero, 2012).

A number of institutions produced data on labour share estimates. The annual macro-
economic database (AMECO) of the European Commission's Directorate General for
Economic and Financial Affairs) provides data based on National Accounts on the
adjusted labour shares for 11 of the G20 countries plus Spain. The labour income share
is calculated as the compensation of employees over total economy GDP multiplied by
total employment. From the two published series market prices and factor costs3 the
long-term downward trend in labour income shares is evident, as can be seen in Figure 1.
When GDP is measured at market prices, the average labour share for the 9 countries
with data from the 1960s declines from a peak of more than 65 per cent to about 56 per
cent in the most recent year (Figure 1, pane A). When measured at factor cost, the
labour share declines from an average of 72 per cent to 63 per cent (Figure 1, panel B).

a miracle (Keynes, 1939) and later Solow questioned the reliability of the empirical evidence (Solow, 1958)
(see La Marca and Lee, 2013).
2 This estimation uses current basic prices. If factor costs are used, the results are similar (see Annex A on

measurement issues). The labour share is defined here as the share of net national income that is received
by workers in the form of labour compensation.
3 AMECO calculates this adjusted labour share with GDP at market prices as well as with GDP at current

factor cost (i.e. minus taxes and plus subsidies). According to Guerriero (2012) the latter is more
meaningful, since taxes do not represent any kind of return to capital or land.

3
 Figure 1. The adjusted labour income share in selected G20 countries and
Spain, estimated by AMECO
Panel A. Market prices Panel B. Factor cost
75

70

65

60

55

50

45
1963
1966
1969
1972
1975
1978
1981
1984

1990
1993
1996
1999
2002
2005
2008
2011
1960

1987

average 12 countries average 11 countries

average 10 countries average 9 countries

Note: The 9 countries are Australia, Canada, Germany, France, Italy, Japan, Spain, the United Kingdom and the United
States. The other series include respectively the Republic of Korea (10 countries), Mexico (11 countries) and Turkey
(12 countries).
Source: AMECO.

The downward trend since the 1990s is also observed by ILO, whose estimates of the
adjusted and unadjusted labour shares are based on main national accounts from UN
Data (Figure 2). The unadjusted labour share is calculated as total compensation over
GDP, and the adjusted labour share increases the unadjusted labour share by the ratio of
self-employed (for a sample of countries for which both series can be constructed
between 1992 and 2011). The period covered is shorter than in the AMECO database,
but includes a larger sample of G20 countries. The adjusted labour shares declines from
an average of about 58 per cent to about 55 per cent, and the unadjusted labour share
falls from 51 per cent to 49 per cent.

Both Figures 1 and 2 show that during the depths of the global economic crisis the
longer term downward trends paused or slightly reversed but began to decline again
following 2009. This reflects the reality that wages tend to be less volatile than profits
during economic downturns. The OECD observed: In times of economic recession,
this decline [in the wage share] has typically paused, but then subsequently resumed with
a recovery. The recent economic and financial crisis and subsequent sluggish recovery
have not deviated from this general pattern (OECD, 2012b, p. 112). Indeed, in most
countries, the labour share tends to increase in the initial years of recessions and then
resume its negative trend afterwards. This countercyclical behaviour of labour shares in
advanced economies has been well documented (see for example the box in IMF, 2012).

4
 Figure 2. The adjusted and unadjusted labour shares in selected G20
countries, estimated by ILO
65.0

60.0

55.0

50.0

45.0

40.0
1994

2001
1991
1992
1993

1995
1996
1997
1998
1999
2000

2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
adjusted labour shares (11 countries + EU)
unadjusted labour shares (11 countries plus EU)
Note: To estimate adjusted labour income shares the projections of number of self-employed from the ILO Global
Employment Trends (GET April 2014 (LP)) are used. Data for the selected group of countries includes EU-28,
Australia, Canada, France, Germany, Italy, Japan, The Rep. of Korea, the Russian Federation, South Africa, United
Kingdom and the United States.
Source: ILO based on main National Accounts from UN DATA (www.data.un.org).

Data using a diversity of sources are used to highlight changes in the labour share at the
country level revealing a decline in the vast majority of countries (Figure 3). The trends
in the aggregate labour share need to be interpreted with some caution due to a number
of measurement issues (see Annex A). For instance, the evolution of the labour share in
the private sector 4 is likely to be shaped by different forces than the corresponding
aggregate for the public sector, where measurement of output and factor shares raises more
complex issues. By contrast, the private represents a consistent aggregate and
measurement issues regarding the labour share are less problematic, at least in OECD
countries.5

4 Refers to those industries where most firms are privately-owned (also referred to as business-sector).
5 In particular, the availability of data on hours worked for both employees and self-employed for a fine
industry partition of the private sector allows refining the imputation of sole proprietors labour earnings,
by assuming their hourly rate in each industry is equal to the average hourly wage in the same industry.

5
 Figure 3. Changes in labour shares in G20 countries (plus Spain)
Panel A. Advanced economies 1970-2014
0.0

-5.0
(per cent)

-10.0

-15.0

-20.0
Spain Italy Republic United Japan Australia Canada Germany France United
of Korea States Kingdom

Panel B. Emerging economies 1995-2012

5.0

0.0

-5.0
(per cent)

-10.0

-15.0

-20.0
Turkey Saudi Arabia Mexico South Africa China India Brazil Russian
Federation

Notes: Figures refer to the change in the adjusted labour share between 1970-2014 for advanced economies and 1995-
2012 for emerging economies. Exceptions include: Republic of Korea: 1991-2014, Saudi Arabia: 2002-09, Turkey and
Mexico: 1995-2014, South Africa: 1995-2013, and Brazil: 1995-2009. Data refer to the adjusted labour income share
except for China and the Russian Federation where the unadjusted labour income share is used. Data for Argentina
and Indonesia are not available. Prior to 1991, the adjusted labour income share in Germany refers to West Germany.
Source: ILO based on AMECO Database and ILO Databases.

Nevertheless, the picture that emerges from focusing on the private sector is rather
similar to the results obtained for the whole economy (Figure 4). The cross-country
average labour share in the private sector, excluding agriculture, mining, fuel and real
estate, was 69.8 per cent in the G20 countries for which data are available in the early
1990s and 65.9 per cent in 2007. On average the contraction over the period was 0.24
percentage points per year. None of the countries for which data are available
experienced a significant trend increase. By contrast, the labour share contracted
significantly in more than three-quarters of the countries. Very large falls in the labour
share were observed in Australia, Canada and Italy where the decline in the private sector

6
labour share exceeded 5 percentage points. The implication is that, in these countries,
labour is obtaining an increasingly smaller share of the priate-sectors pre-tax revenue.
Figure 4. Private-sector labour shares in selected G20 countries, 1990a-2007b
Level in 2007 (a) Level in 1990 (b) Change 1990-2007 (percentage points)
%
5
80

75
0
70

65
***
-5 *** ***
60 *
***

55 *** ***

50 -10

45

40 -15
CAN

DEU
AUS

USA

ITA

FRA
KOR

GBR
JPN

AUS

USA
CAN

DEU
ITA

FRA
GBR

KOR

JPN
Notes: 3-year averages, starting and ending with indicated years. ***,**,* significant at the 1%, 5% and 10% level,
respectively. Statistical significance refers to the coefficient of the time trend in a bivariate regression on annual data
with the labour share as dependent variable. The wage of the self-employed is imputed assuming that in each
industry their hourly wage is the same as for the average employee of the industry.
a) Germany: 1992.
b) Canada: 2004; Korea: 2005; Japan: 2006.
Source: OECD calculations based on OECD STAN and EUKLEMS.

One question arises whether the low level in the labour share witnessed today
corresponds to a historical low point when looking at data over the entire 20th century.
Annex B shows the evolution of the labour share of income using historical data for
France (1897-2010), United Kingdom (1856-2010) and the United States (1899-2010).
The labour share has reached a low point since World War II and is back to its 1897 level
in France. Similarly in the United Kingdom, the labour share is reaching a level similar to
the level reached before World War II. In the United States, the labour share shows a
decline in the past 30 years that pushes the labour share down to a level lower than the
level reached in the 1930s.
(b) Wages and productivity

Trends in labour shares are to a large extent driven by the comparative evolution of
average wages and labour productivity. Under most circumstances, when average wages
increase more rapidly than average labour productivity, the labour share increases.
Conversely, when the growth in average wages lags the growth in labour productivity, the
result is a decline in the labour share. This relationship can become more complex if the
proportion of wages in total compensation changes over time, or if different deflators are
used to deflate wages and output per worker (CPI and GDP deflators can be
substantially different). The ILO (2014) has shown that in a number of countries where
labour shares declined, wage growth significantly lagged behind productivity growth,
even when different deflators are used or if total compensation is used instead of the
narrower concept of wages. Because a majority of large economies, including the United
States, Germany and Japan, have seen wage growth lagging behind productivity growth,

7
labour productivity has outpaced real average wage growth in a group of nine advanced
G20 economies for which data is available since 1999 (Figure 5).
Figure 5. Evolution of average wages and labour productivity in selected
advanced G20 economies, 1999-2013
120
Labour
productivity
index
115

110

Real wage
index
105

100
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
Notes: Data refer to Australia, Canada, France, Germany, Italy, Japan, Rep. of Korea, the United Kingdom and the
United States. Real wage growth is calculated as a weighted average of year-on-year growth in real average monthly
wages in the advanced G20 economies (for a description of the methodology, see ILO Global Wage Report 2014-15,
Appendix I). Index is based on 1999 because of data availability.
Sources: ILO Global Wage Database; ILO Trends Econometric Models, Apr. 2014.

In most of the G20 advanced countries for which data is available, the aggregate growth
of real wages was significantly slower than that of aggregate productivity even taking into
account the dynamics of relative prices, thus accounting for the decline in the labour
share. However, this does not mean that there was slower growth of average real wages
with respect to aggregate productivity within all industries. In fact, in many countries, real
wages grew faster than productivity in a number of industries and less than productivity
in others. Yet, in these countries, the labour share decreased on average within industries
because productivity grew faster than wages in high-productivity industries, thereby
raising the average growth of productivity above that of wages. This phenomenon is
quantified in Annex C.

2. What determines labour shares?

(a) Sectoral shifts and within-sector changes

Most studies that have documented the fall in the labour income share since the 1980s
have also tried to understand its causes. One possibility that has been studied is that
trends in the labour share are determined by a compositional shift in employment from
labour-intensive to more capital-intensive sectors, where labour shares are lower.

8
Indeed, the labour share is significantly different across industries. A key question,
therefore, is whether the decline of the aggregate labour share has been the result of a
structural shift away from labour-intensive activities or whether instead it has been the
result of a decline in the labour share within each industry. Shift-share analysis based on
comparable data for 26 OECD countries and 20 industries in the private sector since
1990, shows that the fall in the labour share is by and large due to within-industry decline
of this share (OECD, 2012). Interestingly, this decline is widespread across industries,
since essentially all industries experienced a considerable decline of the labour share in
the last 20 years. By contrast, although part of the secular decline observed in the labour
share of OECD countries can be attributed to the fact that the economy is shifting away
from agriculture, the contribution of changes in the value-added share of industries with
high labour shares versus those with low labour shares has remained limited. An ILO
study also found that the shift in sectoral composition was indeed a contributory factor
in many countries, but that in advanced economies most of the fall in the labour share
was the result of falling shares within industries (ILO, 2010).
(b) Possible determinants of changes in the labour income share

The usual explanations for changes in labour shares include technological change,
globalization, financial markets, product and labour market institutions, the bargaining
power of labour and unemployment. In this section the paper briefly reviews all the
factors that have been associated with the decline in different studies. There is a wide
diversity of findings in the literature on the relative importance and effects of these
factors, and further evaluation of these different factors will be included in future
revisions of the present paper.

Technological changes are often presented as the main culprit, with some authors seeing
the role of capital accumulation and capital-augmenting technical change as determinants
of the evolution of the labour share (see e.g. Bentolila and Saint-Paul, 2003; Arpaia et al.,
2009; Driver and Muoz-Bugarin, 2010; Raurich et al., 2012; Hutchinson and Persyn,
2012). According to OECD estimates (OECD, 2012), total factor productivity (TFP) growth
and capital deepening the key drivers of economic growth accounted for most of the
average within-industry decline of the labour share in OECD countries between 1990
and 2007 (see the extended discussion in Annex D).

Studies typically also find smaller negative effects of globalization on the labour share in
high-income countries, possibly due to the intensification of competition and the entry
of labour-abundant countries into the global economy which may have worked as a
wage-moderating factor (ILO, 2008). It is also possible that redistribution from labour
to capital has occurred through offshoring or the so-called threat effects that can occur
even without actual changes in production locations (Epstein and Burke, 2001). The role
of financial markets has also been highlighted, particularly its influence on businesses to
increase shareholder value and to focus on their core activities while subcontracting
labour-intensive activities (Weil, 2013; ILO, 2012). Some groups of workers, particularly
top executives, may have benefited from this process of financialization through
deferred salaries in the form of pension funds and other types of capital gains. For the

9
average worker, though, the evidence indicates that the extent and size of such gains are
limited (Bell and Van Reenen, 2013). It has also been observed that capital account
liberalization reforms increase inequality by reducing the labour income share (Furceri et
al., 2015).

Institutional factors have also been explored. Labour market institutions, product
market regulations, the extent of public ownership, the bargaining power of workers and
the size of the welfare state are all among the variables that have been debated in the
existing literature. Of particular attention have been factors such as union density,
minimum wage legislation, unemployment benefits and coverage, severance pay and
government consumption. The decline in union density the number of trade union
members as a percentage of total employees or as a percentage of total employment in
many developed economies has often been linked to the weakening of workers
bargaining power, negatively affecting their ability to negotiate a larger share of
productivity growth as labour compensation. The level of the minimum wage and other
intermediary institutions, including employment protection legislation, the generosity
of unemployment benefit and other benefits and contributions (the tax wedge), are
among the institutional variables that have been widely used in empirical studies (ILO,
2012; IMF, 2007; European Commission, 2007; OECD, 2012). High unemployment can
place downward pressure on wage demands and on the labour share, while the level of
unemployment benefits can have an impact on the labour share by affecting workers
reservation wages, that is, the level of pay workers would accept as a minimum.
Among institutional factors, empirical evidence suggests that the role of factors that
affect the bargaining power of workers is largest (OECD, 2012). As with other potential
determining factors as discussed above, more empirical studies are needed to determine
how labour market institutions affect the evolution of the labour share, particularly
through influencing the bargaining power of workers. Better understanding is also
needed on the interactions between labour market institutions and other markets (e.g.,
finance and product markets) in shaping the trends in the labour share (this issue will be
discussed in depth in the third EWG meeting in July 2015).

3. The effects of declining labour shares

Declining labour shares are frequently associated with more income inequality because
capital is more concentrated than labour endowments. At the same time, the
relationship between factor shares and personal distribution can be relatively complex
because economic agents derive their earnings from several different sources (see
Atkinson, 2009). Some studies have combined factor shares with distributional data.
The ILO (2011) found falling labour shares for low- and medium-skilled workers, but
increasing shares for highly skilled workers in sample of ten developed economies. The
OECD (2012) showed that the labour share diminished for the bottom 99 per cent of
income earners, but increased by 20 per cent for the top 1 per cent in countries for
which data was available over the last two decades (OECD, 2012). Recent work (e.g.
OECD, 2011a, Atkinson et al., 2011) shows that top income earners have seen their share
of national income increase. One study shows that in Europe, the top 10 per cent in the

10
wage distribution earns 25 per cent of the total wage bill while the top 10 per cent in the
capital distribution owns 60 per cent of total capital, so that ultimately - the top 10 per
cent in the distribution of incomes (wages and capital) obtains 35 per cent of national
income. In the U.S., these figures for the top 10 per cent are estimated, respectively, at
35 per cent for wages, 70 per cent for capital, and 50 per cent for incomes (wages and
capital) (Piketty, 2013). Although not demonstrating causality, Figure 6 suggests that the
decline of the labour share tended to evolve hand-in-hand with the widening of market-
income inequalities. Fiscal consolidation in 17 OECD countries over the period 1978-
2009 has also had distributional effects by raising inequality and decreasing labour
income shares (Ball et al., 2013).
Figure 6. Changes in the labour share and in income inequality in OECD
countries, 1990s to mid-2000sa

Change in the Gini coefficient for market income

0.10
ITA Corr. coeff. = -0.57

0.08

NOR
0.06
SWE
USA
0.04 CAN
DEU DNK
CZE
LUX
0.02
GBR
BEL
FRA
0.00
AUS
FIN
GRC
- 0.02

NLD
- 0.04

- 0.06
-10.000 -8.000 -6.000 -4.000 -2.000 0.000 2.000 4.000
Change in the aggregate labour share, percentage points

Notes: Labour share: 3-year moving averages centred around start and end dates. The wage of the self-employed is
imputed assuming that their annual wage is the same as for the average employee of the whole economy. The Gini
coefficient is based on pre-tax and transfer income of the population aged 18 to 65 years.
a) 1990-2004 for Canada; 1990-2005 for Denmark, Netherlands and the United States; 1991-2004 for Italy, Sweden
and the United Kingdom; 1995-2004 for Australia, Belgium, Germany and Norway; 1995-2005 for Finland; 1996-2004
for Czech Republic, France and Luxembourg; 1999-2004 for Greece.
Source: OECD calculations based on the OECD income distribution database, OECD STAN and EUKLEMS .

In addition to affecting the personal distribution of income, labour shares also affect
macroeconomic aggregates. At the level of enterprises, wages represent a cost to
enterprises. At the household level, wages are a significant determinant of household
consumption. At the country level, the sum of all enterprise-level wage changes can have
complex effects. Recent research (Lavoie and Stockhammer, 2013) has shown that a
falling wage share means growing constraints on consumption demands on the one hand,

11
but more exports and potentially more investment on the other. The positive effect on
consumption of redistribution from the capital to the labour share has been attributed to
the fact that the propensity to consume out of labour compensation is higher than the
propensity to consume out of capital income, as the labour share goes to households
with lower incomes and therefore able to save lower proportions of their income than
wealthier people, who save a higher proportion of their total incomes.

It might be argued that lower wages are necessary to boost profits in order to increase
investment and, in turn, job creation. However, in developed economies, the shift in
income away from labour towards capital has not produced the expected results on
investment. Between 2000 and 2007, the capital share in advanced G20 countries grew
by close to 2 percentage points. In contrast, investment as percentage of GDP did not
keep pace and remained stable (from 22.4 per cent in 2000 to 22.8 per cent in 2007).
Since the onset of the global crisis, investment as a percentage of GDP in advanced G20
countries has declined steeply. In 2012, the most recent year with available information,
investment as percentage of GDP was, on average, 20 per cent, almost 3 percentage
points below the peak reached just before the crisis. It is important to note that
investment decreased much more than what had been expected on the basis of the
stability in capital as percentage of GDP (Figure 7, panel A).

However, there is a stark contrast between advanced and emerging countries. Although
in the emerging countries profit shares rose at similar pace than advanced countries,
investment increased by close to 7 percentage points between 2000 and 2007. Moreover,
investment in emerging G20 countries continued to increase over the crisis period, by 2.5
percentage points between 2007 and 2012 (Figure 7, panel B).

There are three main factors explaining the disconnection between growing profits and
productive investment in advanced G20 countries. First, much of the increase in profits
accrued in the financial sector. Secondly, in advanced economies, profits of non-
financial corporations have increasingly been used to pay dividends and to invest in
financial assets rather than to make productive investments. Finally, more recently,
productive investment in advanced economies has been hampered by weak household,
government and trade demand, combined in many countries with tight credit conditions,
affecting small and medium-sized enterprises disproportionately (ILO, 2011).

It is important to note that, when the negative impacts of falling labour share on private
consumption are not offset by investment, countries tend to rely more on credit
(household debts) and/or net exports in order to maintain aggregate demand. This may
contribute to increasing economic instability and global imbalances. If many countries
simultaneously pursue policies of wage moderation (as defined by wage growth lower
than labour productivity growth), the result is likely to be a shortfall in global aggregate
demand, with negative effects on most countries. If this occurs within a trading bloc
with close economic ties, such as the European Union, the results can depress demand
throughout the bloc and beyond.

12
 Figure 7. Capital share and investment developments in G20 countries
(percentages of GDP)
Panel A. Advanced G20 countries
110

105 Capital share

100

95
Investment
(% GDP)
90

85

80
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

Panel B. Emerging G20 countries

135
Investment
130 (% GDP)

125

120

115

110
Capital share
105

100

95
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
Note: The sample analyzed comprises 17 G20 countries, namely Australia, Brazil, Canada, China, European Union,
France, Germany, India, Italy, Japan, Mexico, Republic of Korea, Russian Federation, Saudi Arabia, South Africa,
United Kingdom and United States. Data for Brazil refer to 2000-2009; China: 2000-2011; European Union: 2002-
2012; India: 2000-2011; Mexico: 2003-2012; Republic of Korea: 2000-2011; Russia Federation: 2002-2012; and Saudi
Arabia: 2002-2009. Income groups are based on gross national income (GNI) per capita, according to the World Bank
country classification. High-income countries are those countries with a GNI per capita of USD 12,476 or more.
Averages are weighted based upon 2013 GDP at purchasing power parity.
Source: ILO calculations based on the OECD and UN National Accounts databases and IMF (2014).

13
Annex A. What is the labour share and how is it measured?

The labour share measures the fraction of national income accruing to labour (Krueger,
1999). Although the idea is simple, there are challenges involved in measuring the labour
share. The unadjusted labor share is usually calculated as the ratio of total compensation
of employees wages and salaries before taxes, plus employers social contributions -
over a national product or income aggregate (Luebker, 2007). Regarding the numerator,
issues arise as to who exactly is an employee (should CEOs be included?) and what
should be counted as compensation (should stock options be counted as labour
income?); the denominator can be measured for example as gross national income (GNI)
or gross domestic product (GDP), and it may be measured at market prices or factor
costs. Because measurement of value added is problematic in some sectors (particularly
public administration, where value added in National Accounts is often just the sum of
labour costs), the analysis sometimes focuses on the corporate sector (Karabarbounis
and Neiman, 2013), or a subset of the economy (OECD, 2012).

However measured, the unadjusted labour share is a lower estimate of the true share of
labour income because compensation of employees excludes the income from self-
employment, which is recorded as mixed income in systems of National Accounts and
may thus implicitly be recorded as capital income. Yet, at least part of mixed income
should be seen as return to labour input, and hence as a relevant part of the labour share.
Various methods to adjust the labour share have been tried in the literature. One simple
method of adjustment assumes that two-thirds of mixed income can be attributed to the
labour share; another method is to attribute to the self-employed the same wage as the
average wages of employees; yet another method is to attribute earnings to the self-
employed equal to wages of employees with similar industry- and personal
characteristics6.

While these differences in the way adjustments can be made affect the level of the labour
share, they do not generally affect trends (ILO, 2010; Guerriero, 2012). It is important
however to carefully interpret adjusted and unadjusted labour shares. Structural shifts
from self-employment (like family farming) to wage employment tend to raise the
unadjusted labour share more than the adjusted one. This should be kept in mind,
particularly when looking at trends in emerging and developing countries where the share
of self-employed workers and unincorporated enterprises is larger than in advanced
economies, and where the unadjusted labour income share is thus generally lower than in
more developed countries. Once labour shares are adjusted for self-employment, it is not
obviously the case anymore that labour shares are lower in poorer countries (Gollin,
2002; Guerriero, 2012).

6 See Guerriero, 2012; Gollin, 2002, Arpaia et al., 2009, or Freeman, 2011

14
Annex B. The labour share in historical perspective: France, United
Kingdom and United States
Panel A: France
1.0

0.9

0.8

0.7

0.6
1897 1907 1917 1927 1937 1947 1957 1967 1977 1987 1997 2007
labour share long term fluctuations

Panel B UK
1.0

0.9

0.8

0.7

0.6
1856 1866 1876 1886 1896 1906 1916 1926 1936 1946 1956 1966 1976 1986 1996 2006
labour share long term fluctuations

Panel C: United-States
0.70

0.65

0.60

0.55
1899 1909 1919 1929 1939 1949 1959 1969 1979 1989 1999 2009
labour share long term fluctuations

Note: The long term fluctuations are obtained preforming a frequency analysis using a maximum overlap discrete
wavelet transform. For details about the methodology see Charpe and Bridji (2015).
Source data: Piketty and Zucman (2014) and Groth and Madsen (2013).

15
Annex C

In Figure A.3, percentage changes in the labour share are decomposed into the
contributions of: i) within-industry average growth differences between real wages and
productivity, measured using a common deflator; ii) the reallocation across industries
(towards or away from high-labour-share industries the standard static composition
effect discussed in subsection 1.2 above); and iii) the correlations between growth rates and
levels of real wages and productivity (see OECD, 2012, Box 3.3 for the detailed
methodology). The latter terms capture cross-industry convergence or divergence
patterns of wages and productivity. In fact, if wages diverge that is, larger growth rates
occur in high-wage industries aggregate wage growth will be faster and, ceteris paribus,
the labour share will increase. By contrast, the converse holds for productivity. For
example, when productivity increase faster in high-productivity industries, but the
growth rate of wages is more homogeneous across industries, aggregate real wages grow
less than aggregate productivity, even taking into account the dynamics of different
deflators, and the labour share will decrease.

Figure A.3. The role within- and between-industry changes in productivity and wages in
a b
explaining trends in the private-sector labour share, 1990 -2007

Average annual contributions, in percentages

Difference between within-industry wage and productivity growth rates, adjusted by changes in deflators
Co-variation between wage levels and growth rates
%
Co-variation between productivity levels and growth rates
Changes in industry composition
1.50

1.00

0.50

0.00

-0.50

-1.00

-1.50

Notes: Extended shift-share decomposition of the percentage change of the labour share in the private or business
sector, i.e. those industries where most firms are privately-owned, partitioned in 20 industries, excluding agriculture,
mining, fuel manufacturing and real estate. The wage of the self-employed is imputed assuming that in each industry
their hourly wage is the same as for the average employee of the industry. The difference in the growth rate in
consumption and industry-specific output deflators has been subtracted from the difference in the growth rate of
industry-specific real wages and productivity.
a) Germany: 1992.
b) Canada: 2004; Korea: 2005; Japan: 2006.
Source: OECD calculations based on EUKLEMS.

16
Annex D

According to OECD estimates (OECD, 2012) based on a wide arrays of econometric

methods for the analysis of cross-industry/cross-country/time-series data (including
different types of dynamic GMM estimators), total factor productivity (TFP) growth and
capital deepening the key drivers of economic growth are estimated to jointly
account for as much as 80 per cent of the average within-industry decline of the labour
share in OECD countries between 1990 and 2007. What explains this strong negative
effect of technical change and capital accumulation? One possible explanation has to do
with the diffusion of information and communication technologies (ICT), which in turn
has created opportunities for unprecedented advances in innovation and invention of
new (increasingly cheaper) capital goods and production processes. This has boosted
productivity but also allowed extensive automation of production and high substitution
between capital and labour (see e.g. Greenwood and Jovanovic, 1999; Brynjolfsson and
McAfee, 2011). This conclusion is confirmed by Arpaia et al. (2009) who, using a
structural model, suggest that in the past twenty years, technical change was in fact
capital-augmenting (while it was labour-augmenting, thereby pushing up the labour share, in
the first half of the twentieth century). By contrast, they argue that the high degree of
substitution between capital and labour was in fact due to high substitution between
capital and low-skilled labour and complementarity between capital and high-skilled
labour. Other scholars have advanced the possibility that, within this context, technical
change could be even labour-replacing, in the sense that technological progress takes the
form of machines replacing tasks previously performed by labour. In turn, this would
especially reduce job opportunities for low-educated workers and, in practice, dampen
the aggregate productivity of low-skilled labour (see Zeira, 1998; Arthur, 2011; and the
survey on machine-replacing-labour technical change in Acemoglu, 2011).

Both interpretations appear consistent with two additional pieces of empirical evidence
emerging from OECD research (OECD, 2012). First, in the last decades, labour
productivity growth has been associated with increases in the share of those with tertiary
education in labour compensation and contractions of the shares of those with lower
levels of education, and particularly those with less than upper secondary education.
Second, decomposing further the association between productivity growth and the
decline of the share of the low-educated in total compensation, ICT capital accumulation
appears to have had an especially negative effect on the lowest educated, while TFP
growth has affected particularly the share of those with intermediate education. These
two results taken together suggest that, in the period under analysis, technical change
embodied in ICT capital was strongly biased against the low-educated, while disembodied
technical change was strongly biased towards high-skilled labour. While the first result is
fully consistent with the literature on skill-biased technical change, one possible
explanation of the latter is that disembodied technical progress reflects embodiment in
intangible capital (entrepreneurship, output from R&D departments, better management,
high-performing human resource practices) that is improvements that are essentially
incorporated in highly-qualified personnel.

17
From a policy perspective, however, it is not possible with the available data to assess
whether the negative relationship between technical progress and changes in the labour
share is a long-lasting relationship or is specific to the past decades and will progressively
disappear when the process of diffusion of ICT-based technologies slows down. On the
one hand, the standard view in the theory of economic growth is that, in the long-run,
capital and labour are complements and technical change augments the factor that
cannot be accumulated (that is labour, see e.g. Acemoglu, 2002). Hence,
capital-augmenting technical change and substitutability between capital and labour are
likely to be only a temporary phenomenon due to the rapid diffusion of ICT-based
technologies and related innovations. By contrast, within this view, to the extent that the
skilled labour supply increases faster, thereby increasing incentives to create capital goods
complementary to skilled labour, technical change would remain biased against the
unskilled. On the other hand, a more pessimistic view considers that ICT has changed
the nature of technological advances, making them more rapid but incorporated in
machines whose main purpose is to replace jobs previously held by certain categories of
workers (Brynjolfsson and McAfee, 2011; Acemoglu, 2011). If this were the case, most
workers, and in particular the least educated, would find themselves in a race against the
machine, thereby increasingly worsening their relative position.

18
References

Acemoglu, D. 2002. Directed Technical Change, Review of Economic Studies, Vol.

69, pp. 781-810.

Acemoglu, D. 2011. When Does Labor Scarcity Encourage Innovation?, Journal of

Political Economy, Vol. 118, No. 6, pp. 1037-1078.

Arpaia, A., E. Prez and K. Pichelmann. 2009. Understanding Labour Income Share
Dynamics in Europe, European Economy Economic Papers 379, May, European
Commission, Brussels.

Arthur, W.B. 2011. The Second Economy, McKinsey Quarterly, October.

Atkinson, A. 2009. Factor shares: The principal problem of political economy? in

Oxford Review of Economic Policy, Vol. 25, No. 1, pp. 316.

Atkinson, A.B, T. Piketty and E. Saez. 2011., Top Incomes in the Long Run of History,
Journal of Economic Literature, Vol. 49, No. 1, pp. 3-71.

Bank for International Settlements (BIS). 2006. 76th Annual Report. Basel.

Ball, L., Furceri, D., Leigh, D., and Loungani, P. 2013. The distributional effects of fiscal
consolidation. IMF Working Paper, Washington.

Bell, B.D.; Van Reenen, J. 2013. Extreme wage inequality: Pay at the very top, in
American Economic Review: Papers and Proceedings, Vol. 103, No. 3, pp. 15357.

Bentolila, S. and Saint-Paul, G. 2003. Explaining Movements in the Labor Share,

Contributions to Macroeconomics, Vol. 3, No. 1.

Bridji S. and Charpe M. 2015. "Income Distribution and Growth using Historical
Data", mimeo International Labour Organization.

Brynjolfsson, E. and A. McAfee. 2011. Race Against the Machine, Digital Frontier
Press, Lexington, Mass.

Driver, C., and J. Muoz-Bugarn. 2010. Capital Investment and Unemployment in

Europe: Neutrality or not?, Journal of Macroeconomics, Vol. 32, pp. 492-496.

Epstein, G. and Burke, S. 2001. Threat effects and the internationalization of

production, Political Economy Research Institute Working Paper 15 (Amherst,
University of Massachusetts).

19
European Commission (EC). 2007. The labour income share in the European Union,
in Employment in Europe 2007, Directorate-General for Employment, Social Affairs
and Equal Opportunities (Brussels), pp. 23772.

Freeman, R. 2011. Accounting for the Self-Employed in labour Share Estimates: The
Case of the United States, OECD Science, technology, and Industry Working Papers,
2011/04, OECD.

Furceri, D., Jaumotte, F., and Loungani, P. 2013. The Distributional Effects of Capital
Account Liberalization. forthcoming IMF Working Paper, Washington.

Glyn, A. 2009, Functional distribution and inequality, in W. Salverda, B. Nolan and

T. Smeeding (eds): The Oxford handbook of economic inequality (Oxford, Oxford
University Press), pp. 10126.

Gollin, D. 2002. Getting income shares right, in Journal of Political Economy

(Chicago), Vol. 110, No. 2, pp. 45874.

Gollin, D. 2008. "Labour's Share of Income," Department of Economics Working

Papers 2008-19, Department of Economics, Williams College.

Greenwood, J. and B.J. Jovanovic (1999), Accounting for Growth, in: C. Hulten (ed.),
Studies in Income and Wealth: New Directions in Productivity Analysis, Chicago:
University of Chicago Press for NBER.

Guerriero, M. 2012. The Labour Share of Income around the World. Evidence from
a Panel Dataset, IDPM, Development Economics and Public Policy Working Paper
Series WP No.32/2012.

Guha, K. 2007. A global outlook, in Financial Times, 17 Sep.

Harrison, A. 2002. Has globalization eroded labors share? Some cross-country

evidence, mimeo, University of California at Berkeley.

Husson, M. 2010. Le partage de la valeur ajoute en Europe, in La Revue de lIRES,

Vol. 64, No.1, pp. 4791.

Hutchinson, J. and D. Persyn. 2012. Globalisation, concentration and footloose

Firms: in search of the main cause of the declining labour share, Review of World
Economics, 148(1).

International Labour Office (ILO). 2008. Global Wage Report 2008/09: Minimum
wages and collective bargaining Towards policy coherence (Geneva).

20
. 2010. Global Wage Report 2010/11: Wage policies in times of crisis (Geneva).

. 2011. World of Work Report 2011: Making markets work for jobs (Geneva).

. 2012. Global Wage Report 2012/13: Wages and equitable growth (Geneva).

. 2014. Global Wage Report 2014/15: Wages and Income Inequality (Geneva).

International Monetary Fund (IMF). 2007. The globalization of labor, in World

Economic Outlook, April 2007: Spillovers and cycles in the world economy
(Washington, DC), pp. 16192

. 2012. World Economic Outlook: Growth Resuming, Dangers Remain. Washington,

April, 2012.

. 2014. World Economic Outlook October 2014: Legacies, Clouds, Uncertainties

(IMF: Washington, D.C).

Jayadev, A. 2007. Capital account openness and the labour share of income, in
Cambridge Journal of Economics, Vol. 31, No. 3, pp. 42343.

Karabarbounis L. and B. Neiman 2014. The Global Decline of the Labor Share, in
The Quarterly Journal of Economics, Vol. 129, No. 1, pp. 61-103

Krueger, A. 1999. Measuring labors share, in American Economic Review, Vol. 89,
No. 2, pp. 4551.

La Marca, M.; Lee, S. 2013. Wages and growth in open economies: A policy
dilemma?, in I. Islam and D. Kucera (eds), Beyond Macroeconomic Stability:
Structural Transformation and Inclusive Development, ILO and Palgrave Macmillan,
2013.

Lavoie, M.; Stockhammer, E. (eds). 2013. Wage-led growth: An equitable strategy for
economic recovery, Geneva and Basingstoke, ILO and Palgrave Macmillan.

Luebker, M. 2007, Labour shares ILO Technical Brief No. 1, Geneva: ILO.

Mankiw, N.G. 2003. Macroeconomics, 5th edn (New York, Worth Publishers).

21
Organisation for Economic Co-operation and Development (OECD). 2011a, Divided
We Stand, OECD Publishing, Paris.

. 2011b, Employment Outlook, OECD Publishing, Paris.

. 2012. Employment Outlook 2012 (Paris).

. 2014. Employment Outlook 2014 (Paris).

Piketty, T. 2013. Le Capital au XXIe sicle (Paris: Seuil).

Raurich, X., H. Sala and V. Sorolla (2012), Factor shares, the price markup, and the
elasticity of substitution between capital and labor, Journal of Macroeconomics, Vol.
34, No. 1, pp. 181-198.

Weil, D. 2013. The Fissured Workplace: why work became so bad for so many and
what can be done to improve it, Harvard University Press.

Wolf, M. 2014. The shifts and the shocks: What weve learned and have still to
learn from the financial crisis (London, Penguin).

Zeira, J. (1998) Workers, Machines and Economic Growth, Quarterly Journal of

Economics, Vol. 113, No. 4, pp. 1091-1113.

22